Research On Reliability Analysis Method Of Slope Stability And Its Factors Sensitivity

The properties parameters of slope soil are uncertainty and correlation, while traditional deterministic analysis methods can no longer be evaluate slope stability comprehensively. So it is important to find an effective method to evaluate slope stability. For that reason the reliable analysis has been an important trend to evaluate slope stability upon deterministic analysis. The paper mainly studied the response surface method of reliability analysis and the sensibility of soil parameters.First, combining with the soil failure mechanism of slope, establish an expression of quadratic polynomial response surface method which suited to slope engineering. Use MATLAB writes a program to analysis the soil slope’s reliable, and the result is very close to Monte-Carlo’s which error is fewer than 5 percent. It’s verifying the feasibility of this method to evaluate the reliable of slope stability.Secondly, the paper analysis the sensibility of soil parameters influences the stability of slope which is important to simplify the reliability analysis. The analysis results show that, the cohesion of soil is most sensitive factors to affect the reliability of slope while the unit weight is least, internal friction angle in between. C-Phi inverse correlation has an advantage to the stability of slope, so we can consider it appropriately when analysis the reliability.Finally, discuss the feasibility of use BP neural network response surface method to analysis the stability of slope for the reason of shortcomings of the quadratic polynomial response surface method. The BP neural network is used to simulate the response surface of slope stability state. And then calculate the reliability index and failure probability of slope which use optimization algorithm. Through homogeneous and heterogeneous soil example analysis show that this method is feasibility and efficient to analysis the reliability of slope.